This was my final TOK (Theory of Knowledge) essay i had to write for my IB diploma...hope you enjoy reading it as much a...
Statistics can be very helpful in providing a powerful interpretation interpretation of reality but also can be used to distort our understanding. Discuss some of the ways in which statistics can be used or misused in different Areas of Knowledge to assist and mislead us, and how we can determine whether to to accept the statistical evidence that is presented to us.
International School of The Hague Name: Paul van Eeghen Candidate number: 000416-094 Word count: 1,571 Date: 23 Apr. 08
‘83% of statistics are made up’ this is meant to be as a joke but if one looks deep deeper er into into the the mean meanin ing g of the the joke joke you you real realiz izee that that it migh mightt be a pret pretty ty good good representation regarding the reliability of statistics that you hear through out your daily life. One could say that statistics cannot be used as pure facts or evidence for something, but rather give a probable answer to the question asked, only until examining the raw data and how it was gathered can the viewer tell if the statistic is reliable or not. This is where society, some might claim, has abused such powers of showing viewers, for example, television advertisements which claim some sort of statistic to provide a good reason to buy that particular product. Whilst people selling the product do not show any sort of raw data or how they might have collected the information they represent. It is safe to assume that this type of statistics statistics could and should be ignored. ignored. As stated already, already, knowing knowing how and what kind of results one gets can help the validity of the statistics shown, of course considering the raw data was not made up. Statistics are often used instead of words such as some, lots and many, for they provide a more accurate answer and does not allow multiple. This essay shall focus on whether statistics are misleading interpretations of the world and to which Areas of Knowledge this applies.
If one looks at statistics then it is important to note in which sector the statistic belongs, such as: social sciences, natural sciences, history. If the statistic is based on social science then it is more likely to be more open to interpretation than if it were based on a natural science. The reason for this is that social sciences deal with, amongst others, human emotions, reason and perception. Whereas natural science deals with physical
000416-094
Paul van Eeghen
2
phenomena, unchangeable by humans, which do not have a certain view point on the research question. Math Mathema emati tics cs allo allows ws one one to calc calcul ulat atee the the stan standar dard d devia deviati tion on and and mean mean of a statistical data. It also allows one to represent a statistical data as a fraction or percentage. If the statistical data is represented in a fraction it is less likely that one will fully comprehend the significance of it, but if it is represented in a percentage then it is easier for one to understand the meaning of it. For example, ‘thirteen out of twenty-three people in the USA dislike Mr. Bush’ one is less likely to understand the amount of people that dislike him. Whilst if it were ‘57% of US citizens dislike Mr. Bush’ people are more likely to understand the significance. Paradigms Paradigms play a great role in ones society and therefore can play a great role in the validity of the statistic to the person. Paradigms have been drilled into the viewers head since birth. If a certain statistic suits the paradigm then you are more likely to believe it then if it does not. For example, ‘one in four women has been raped’1 if this statistic fits the paradigm you has been brought up with and lived your whole life with then it is more likely that you will believe such a claim and support it without any doubt or question of its validity. But if the claim does not fit the paradigm and you have a different opinion on it, then it is less likely that that person will want to believe it and therefore will want to challenge such a claim and may require more evidence, such as the way the statistic was gathered and/or the validity of the statistic. There are different ways in which a statistic can be formulated, such as: response bias, the syntax of the question and/or the diction of the question. Using one or all of the
1
http://www.spiked-online.com http://www .spiked-online.com/Articles/00000 /Articles/0000000054A4.htm 00054A4.htm 05 Feb. 2006
000416-094
Paul van Eeghen
3
different ways can lead the person that is answering, to respond in a specific way the questioner questioner wants him/her to reply. reply. A response response biased biased2 question involves getting the person that answers to answer in a certain way which the questioner already knows the answer for. For example, if the European public was shown short video extracts of the war in Iraq where people are screaming in the streets for help and civilians being shot and they were asked if the involvement there should be continued then most likely the majority would answer that they should not continue the involvement there. But if the same question is asked but short video extracts are shown of people’s freedom and right to go to school then the majority would answer that it is vital for them to stay there and protect their freedom. The syntax of the question can also lead one to answer in a certain way the questioner wants. For example, if a flight attendant asked the business section of the plane a simple question like ‘For your main course would you like beef or chicken?’3 if it is a person with no real preference or need for a specific meat, then most would answer that they would like ‘chicken’ as that is the last word they remember from the question. But if the syntax was changed to ‘For your main course would you like chicken or beef?’ and there is no preference then most would answer ‘beef’ for the same reason. The diction of a question is also very important whether a serious answer is required or not. If the diction is very informal and colloquial then one will most probably not take the question too seriously and therefore answer in an informal manner. But on the other hand, h and, if the question contains a lot of formal and complex words then the person that is answering might forget/lose the meaning of the question and therefore answer in the manner that ones brain remembers the question, which might be completely different to
2 3
http://en.wikipedia.org/wiki/Response_bias http://en.wikipedia.org/wi ki/Response_bias 06 Mar. 2006 TOK class discussion
000416-094
Paul van Eeghen
4
his paradigms and the actual question. The people the questioner chooses to ask can also make the statistic more reliable or not. There are several ways in which the questioner can choose people such as: stratified sampling, random sampling, quota sampling and cluster sampling.4 Stratified sampling is when the questioner chooses to ask ten women and ten men but the people that are chosen are random (rich/poor/athletic). This could be a good way of gathering raw data as there is an equal amount of people represented but random people are chosen. Random sampling is when the questioner asks any random person they meet. If the equality of representatives, such as male and female, is irrelevant then this would be the best way. Quota sampling is when an equal amount of representatives are asked and the peop people le asked asked are are also also chos chosen en.. This This is wher wheree the the vali validi dity ty of the the stat statis isti ticc could could be questioned as the people could have been chosen specifically to support a certain view point. Cluster sampling is when the people asked are chosen by geographical location. This can be a good representation of a governmental party’s chance of getting elected as the question is asked relative to location rather than status. Sources can also be a powerful way of representing data. Sources can be in different forms like: a friend telling you a statistical fact; a webpage; professor telling a statistical fact and television advertisements. For example, if a friend tells you a certain statistic such as ‘there are 5 * 10 23 stars in the universe’ and this does not fit your paradigm then you are unlikely to believe your friend. But if you see a professor on television in a lab coat with three pens, black, blue and red, in his left pocket says that there are 5 *10 21 5 stars in the universe then you are more likely to believe the stranger on 4 5
http://www.bized.ac.uk/learn/business/market http://www.bized.ac.uk/lear n/business/marketing/research/inde ing/research/index.htm x.htm 05 Feb. 2006 http://www.everything2.com http://www .everything2.com/index.pl?node_id= /index.pl?node_id=681430 681430 05 Feb. 2006
000416-094
Paul van Eeghen
5
television rather than the friend. If one was told that webpage sources are unreliable and you read that there are 5 * 10 21 stars in the universe and later on reads in a newspaper that there are 5 *10 23 stars in the universe then he is more likely to believe the newspaper’s claim then the webpage’s claim. Television advertisements can be very influential for naive viewers as they will most likely believe the claims they hear on television. For example, if a person on television asks random people on the streets what they think of the new product then they could edit out the negative remarks made by people and therefore make the product look like it is very good.
Therefore statistics can be very powerful tools for p oliticians and within the social sciences as they influence the way the person asked interprets the question and therefore the answer. The paradigms of ones generation also influence the validity, as they can affect to what degree one needs to see the raw data or the question asked. It can also influence ones interpretation on what to do about it, whether it is to vote for a certain party or to take action against the claim. This is a reason one might be pessimistic on whether the statistics shown are to be trusted. This is less likely the case with natural sciences as they do not deal with the human emotions but rather deal with nature’s reaction to a certain situation. Although both social and natural sciences deal with the viewer’s interpretation of the statistic, I believe a statistic from a natural science more reliable than statistics that come from the human sciences.
000416-094
Paul van Eeghen
6
Bibliography •
Iain. There are more stars in the universe than g rains of sand on
[email protected].. 02 Aug. 2000
[email protected] > 05 Feb. 2006
•
Josie Appleton. spiked-central | Feature | Statistical abuse. 13 Feb. 2001 05 Feb. 2006
•
Market Research - Marketing - Business Studies - Learning Materials. 05 Feb. 2006 05 Feb. 2006
•
Response bias - Wikipedia, the free encyclopedia. 12 Feb. 2006 > 06 Mar. 2006
000416-094
Paul van Eeghen
7